中文摘要：

利用QR分解的稳定性以及提升小波计算速度快的优良特性,给出一种基于QR分解和提升小波变换的盲鲁棒数字音频水印方法.为了保护原始二值水印图像的安全,利用混沌序列对其进行扩频,生成了待嵌入的水印信号.将原始宿主音频信号升维后进行QR分解,根据R分量是上三角矩阵且第一行为非零元素的特点,选定R分量的第一行,对其进行提升小波变换,得到了待嵌入的小波系数,利用线性瞬时混合模型将其与待嵌入的水印信号进行混合,得到隐秘音频信号.水印提取时,利用独立分量分析算法从待检测的隐秘音频信号中提取嵌入水印信号,获得嵌入水印信号的估计,经过后处理即可获得水印图像.实验结果表明,该方法可以实现水印的盲提取,并且具有良好的透明性和鲁棒性.

英文摘要：

Making use of the good stability of QR decomposition and good computing ability of lifting wavelet,a blind robust digital audio watermarking method based on QR decomposition and lifting wavelet transform is proposed. To protect the security of the original binary watermarking image,chaotic sequence is used to spread its spectrum and watermarking signal for embedding is generated. In the watermark embedding procedure,first,QR decomposition is executed after rising dimension of the original host audio signal; then,as R component is an upside-triangle matrix and the element in its first row is nonzero,lifting wavelet transform is taken on its first row to obtain the selected wavelet coefficients; last,the watermarking signal is embeded into the selected wavelet coefficients using linear instantaneous mixing model to generate the watermarked audio signal. In the watermark extraction procedure,independent component analysis algorithm is utilized to extract the watermarking signal from the watermarked audio signal for detecting. The watermarking image can be obtained after post-processing. Experimental results show that this method can achieve blind extraction of watermark and has good transparency and robustness.